This Article Abstract Full Text (PDF) Supplemental Data All Versions of this Article: 18/6/1477    most recent tpc.105.040394v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (12) Citing Articles via Google Scholar Google Scholar Articles by daSilva, L. L.P. Articles by Denecke, J. Search for Related Content PubMed PubMed Citation Articles by daSilva, L. L.P. Articles by Denecke, J. Agricola Articles by daSilva, L. L.P. Articles by Denecke, J.

Targeting of the Plant Vacuolar Sorting Receptor BP80 Is Dependent on Multiple Sorting Signals in the Cytosolic Tail^[W]

Centre for Plant Sciences, Faculty of Biological Sciences, University of Leeds, Leeds LS2 9JT, United Kingdom

¹ To whom correspondence should be addressed. E-mail j.denecke{at}leeds.ac.uk; fax 44-113-3432835.

	ABSTRACT

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Although signals for vacuolar sorting of soluble proteins are well described, we have yet to learn how the plant vacuolar sorting receptor BP80 reaches its correct destination and recycles. To shed light on receptor targeting, we used an in vivo competition assay in which a truncated receptor (green fluorescent protein-BP80) specifically competes with sorting machinery and causes hypersecretion of BP80-ligands from tobacco (Nicotiana tabacum) leaf protoplasts. We show that both the transmembrane domain and the cytosolic tail of BP80 contain information necessary for efficient progress to the prevacuolar compartment (PVC). Furthermore, the tail must be exposed on the correct membrane surface to compete with sorting machinery. Mutational analysis of conserved residues revealed that multiple sequence motifs are necessary for competition, one of which is a typical Tyr-based motif (YXX). Substitution of Tyr-612 for Ala causes partial retention in the Golgi apparatus, mistargeting to the plasma membrane (PM), and slower progress to the PVC. A role in Golgi-to-PVC transport was confirmed by generating the corresponding mutation on full-length BP80. The mutant receptor was partially mistargeted to the PM and induced the secretion of a coexpressed BP80-ligand. Further mutants indicate that the cytosolic tail is likely to contain other information besides the YXX motif, possibly for endoplasmic reticulum export, endocytosis from the PM, and PVC-to-Golgi recycling.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
In contrast with soluble proteins that are delivered to the extracellular matrix by default, soluble and membrane proteins designated to a specific intracellular compartment require information for their sorting and/or retention (Jurgens, 2004). The Golgi apparatus is the site where a large number of decisions are made regarding the transport of proteins in the secretory pathway, and the vacuole/lysosome is one of the target compartments that can be reached from this organelle.

In plants, vacuolar transport appears to be particularly complex due to the wide range of functions performed by vacuoles (Marty, 1999) and the existence of at least two distinct types of vacuolar compartments (Robinson et al., 2005). In addition to the central lytic vacuole, thought to be the equivalent of the yeast vacuole and the mammalian lysosome, protein storage vacuoles were shown to coexist in the same cells (Hohl et al., 1996; Paris et al., 1996; Bassham and Raikhel, 2000). To maintain their highly distinct protein composition, these compartments need to be efficiently distinguished by mechanisms of cargo segregation and subsequent targeting. Sorting to the lytic vacuole is Golgi-mediated and dependent on a type I transmembrane protein, originally termed BP80 (80-kD binding protein), which was first isolated from clathrin-coated vesicle extracts of pea (Pisum sativum) cotyledons (Kirsch et al., 1994). Recently, a related membrane spanning protein with a shorter lumenal domain and a longer cytosolic tail (CT) bearing a RING-H2 domain (RMR) was proposed to carry out a similar function for Golgi-mediated sorting to the storage vacuole (Jiang et al., 2000; Park et al., 2005).

Unlike RMR, which accompanies its ligands to the storage vacuole (Jiang et al., 2000; Park et al., 2005), BP80 is thought to recycle from the prevacuolar compartment (PVC) to the Golgi to avoid degradation in the lytic vacuole. This is proposed to allow renewed cargo collection at the Golgi membrane, similar to vacuolar sorting receptors in yeast and lysosomal sorting receptors in mammalian cells (Seaman, 2005). In support of this model, BP80 was shown to localize to the trans-Golgi network and the PVC (Paris et al., 1997; Sanderfoot et al., 1998; Li et al., 2002). The latter organelle often contains electron-dense internal vesicles and is also termed the multivesicular body (Tse et al., 2004). To function according to the model, BP80 must specifically interact with at least two different types of cytosolic components needed for vesicle traffic. One is required for cargo-loaded anterograde traffic from the Golgi membrane, whereas the second is needed for retrograde recycling from the PVC after cargo release. BP80 must bind ligands in the Golgi apparatus, release them in the PVC, and ensure that it is recycled without the ligand. How these intricate steps are accomplished in vivo is not understood.

Although BP80 was first postulated as a specific receptor for lytic vacuolar cargo, it has been suggested that BP80-related receptors could also mediate transport to the storage vacuole (Shimada et al., 2003; Jolliffe et al., 2004; Vitale and Hinz, 2005). Considering the premise that plants have two distinct types of vacuoles within the same cells (Paris et al., 1996), different ligand binding specificities of the lumenal portion of BP80 would be required to distinguish storage proteins from hydrolases destined to the lytic vacuole. Equally important, a storage vacuole–specific BP80 should exhibit differences in the CT to recognize distinct adaptor molecules and coat components.

On first sight, BP80 sorting appears to occur in a similar fashion as in mammalian cells and yeast, where active cargo selection is mediated by type I membrane spanning proteins that contain a lumenal domain to recruit ligands and a cytosolic domain to form clathrin-coated vesicles for transport to the PVC or prelysosome (Marcusson et al., 1994; Ghosh et al., 2003). Likewise, the plant PVC is thought to be the equivalent of the late endosome in mammals and yeast, which serves to rescue sorting receptors from the lysosomes or vacuoles. This occurs by transport via the retromer complex back to the Golgi apparatus, and receptors are thus recycled for new rounds of cargo selection (Seaman et al., 1997, 1998; Seaman, 2005).

A highly conserved Tyr motif (YMPL/YIPL), present in all Arabidopsis thaliana BP80 isoforms (Hadlington and Denecke, 2000), was shown to be important for in vitro interaction with the µ-subunit of mammalian AP1 (Sanderfoot et al., 1998) and to a Golgi-localized µ-adaptin from Arabidopsis (Happel et al., 2004). Together with the initial purification of BP80 from a clathrin-coated vesicle–enriched fraction (Kirsch et al., 1994), these findings suggest that BP80 recruits adaptor protein (AP) complexes to the Golgi membranes and integrates clathrin coat assembly with its inclusion into the nascent vesicle (Robinson et al., 2005). These reports appear to suggest an active role for the BP80 CT in the clathrin-mediated anterograde transport from the Golgi apparatus to the PVC.

Surprisingly, it was shown that the transmembrane domain (TMD) of BP80 without the CT was sufficient for targeting of a hybrid type I membrane protein to a lytic compartment (Jiang and Rogers, 1998). In addition, the tonoplast was proposed to be the default destination for transmembrane proteins in plants (Barrieu and Chrispeels, 1999). However, the length of the membrane-spanning domain was shown to control the final destination of transmembrane proteins on the plant secretory pathway, and in the case of BP80, this was localized to the Golgi apparatus when no cytosolic domains were present (Brandizzi et al., 2002b). The discrepancies among the published findings call for a thorough investigation into the domains responsible for BP80 sorting in vivo.

We have recently introduced a novel approach to monitor receptor traffic in vivo using a truncated BP80 molecule that permits in vivo imaging and interferes with the sorting of endogenous BP80 (daSilva et al., 2005). A chimeric protein in which the TMD and CT of BP80 were fused to the C terminus of secreted green fluorescent protein (GFP-BP80) or secreted yellow fluorescent protein (YFP-BP80) is able to use receptor transport machinery to reach the PVC (Tse et al., 2004; daSilva et al., 2005). Due to this property, it competes with endogenous receptors, particularly at the retrograde transport step from the PVC, and causes receptor leakage to the vacuole and depletion in the Golgi apparatus. This manifests itself by the strongly induced secretion of BP80-ligands, which is easy to quantify (daSilva et al., 2005), and permits the monitoring of BP80-related sorting events by in vivo imaging and via cargo transport assays. It is also possible to reconstitute vacuolar sorting in the presence of constant GFP-BP80 levels by reintroducing full-length BP80 molecules (daSilva et al., 2005). The use of these tools allowed us to demonstrate that recycling of BP80 from the PVC to the Golgi is crucial in maintaining receptor function and that this step is wortmannin-sensitive (daSilva et al., 2005).

Here, a combination of these complementary assays was further explored and expanded to identify sorting determinants of BP80. We could show that both the TMD and the CT play a role in proper sorting, including endoplasmic reticulum (ER) export, Golgi-to-PVC transport, and recycling. In addition, a Tyr-based motif in the CT was shown to be important in vivo for efficient Golgi-to-PVC transport but not essential for ultimately reaching the PVC due to the presence of an alternative pathway via the plasma membrane.

	RESULTS

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
A Role for the CT of BP80 in Receptor Sorting
Two studies have addressed the role of the TMD and the CT of BP80, with contradictory findings (Jiang and Rogers, 1998; Brandizzi et al., 2002b). To directly test if the CT contains targeting information, we have generated a derivative of the competitor GFP-BP80, in which the entire CT was deleted (Figure 1A ; GFP-BP80CT). We have previously shown that the arrival of GFP-BP80 in the lytic vacuole can be monitored by the detection of a characteristic GFP degradation product, termed the GFP-core (daSilva et al., 2005). In addition, glycosylated GFP is deglycosylated in transit to the vacuole. Therefore, GFP-BP80 generally gives rise to three polypeptides that can be detected: the high molecular weight glycosylated precursor, the deglycosylated precursor, and the low molecular weight GFP-core.

View larger version (29K): [in this window] [in a new window]   Figure 1. Competition with the Endogenous Receptor Depends on Membrane-Anchored CT. (A) Schematic representation comparing the topology of the additional BP80-truncated fusion proteins in relation to full-length BP80 and the previously characterized competitor in which the complete lumenal ligand binding domain of Arabidopsis BP80 isoform a (Hadlington and Denecke, 2000) was replaced by GFP (GFP-BP80; daSilva et al., 2005). The entire sequence encoding the BP80 CT in GFP-BP80 was deleted, yielding the tailless protein GFP-BP80CT. The cytosolic protein cGFP-CT was generated by fusion of the CT of the same BP80 isoform to the C-terminal end of the GFP sequence lacking the signal peptide for ER translocation. (B) Transient expression experiment to compare the intracellular partitioning of the three BP80 chimeras. Tobacco mesophyll protoplasts were mock transfected (con) or transfected with 30 µg of plasmid encoding either GFP-BP80, GFP-BP80CT, or cGFP-CT and incubated for 24 h, after which medium and cells were harvested. Cells were fractionated to obtain extracts enriched in soluble proteins released by osmotic shock (S1), proteins released after sonicating the first pellet (S2), and membrane spanning proteins (P). These cell fractions together with the medium were analyzed by protein gel blotting and probed with anti-GFP antibodies. Molecular weight markers are indicated in kilodaltons. Notice that the vacuolar degradation product, the GFP-core (asterisk), is present in lower amounts in the case of GFP-BP80CT compared with GFP-BP80, and it is absent for cGFP-CT. (C) Influence of GFP-BP80, GFP-BP80CT, or cGFP-CT on the secretion index of amy-spo. Protoplasts were transfected with a constant amount (10 µg) of plasmid-encoded amy-spo alone (Con) or cotransfected with increasing concentrations of plasmid encoding either of the three BP80 chimeras. Plasmid concentrations are shown below each lane. The secretion index was calculated as the ratio between the extracellular and intracellular -amylase activities (Phillipson et al., 2001). Standard errors are from four independent protoplast transfections. Notice that in sharp contrast with GFP-BP80, neither GFP-BP80CT nor cGFP-CT induces secretion of amy-spo.

We then performed the competition assay using an -amylase fusion protein carrying the sorting signal of sweet potato (Ipomoea batatas) sporamin (amy-spo), which is an established BP80-ligand and vacuolar cargo (Pimpl et al., 2003; daSilva et al., 2005). Tobacco (Nicotiana tabacum) leaf protoplasts were transfected with equal amounts of amy-spo encoding plasmids and an increasing concentration of either GFP-BP80 or GFP-BP80CT plasmids. After 24 h, medium and cells were harvested, and the -amylase activity was measured for each fraction as the change in optical density in function of time and volume of the cell fraction used. This allowed us to calculate the secretion index, which is the ratio between the activity in the medium and in the cells (Phillipson et al., 2001; Pimpl et al., 2003).

Confirming our previous results (daSilva et al., 2005), GFP-BP80 strongly induced the secretion of amy-spo in a dosage-dependent manner (Figure 1C). However, the fusion protein lacking the CT failed to exhibit any measurable effect regardless of the concentration. Comparable levels of the two effector molecules were produced from each plasmid (Figure 1B). This result illustrates that the CT is crucially necessary for the ability of the chimera to interfere with endogenous BP80 function and that induced secretion is not an indirect effect of a coexpressed type I membrane spanning protein.

Although we have shown previously that competition mainly involves the retrograde route of the receptor (daSilva et al., 2005), it is not clear whether this is due to titration of limiting cytosolic components or to overload of transport carriers. To distinguish between these two possibilities, we fused the CT alone to the C terminus of cytosolic GFP (Figure 1A; cGFP-CT) where it would have access to cytosolic machinery components such as adaptor molecules. The resulting protein is highly expressed in the cytosol and is not proteolytically trimmed to yield the core fragment (Figure 1B). To test if the resulting CT can titrate sorting machinery, we repeated the competition experiment using amy-spo as a marker. Figure 1C shows no detectable competing activity on BP80-mediated sorting of amy-spo, even in the highest concentration. This shows that the CT alone is not sufficient to recruit sorting machinery. Probably, the tail must be membrane anchored to mediate effective competition with the endogenous receptor.

The CT of BP80 Is Required for Efficient Anterograde Transport
To investigate the influence of the CT on the kinetics of proteolysis, we performed pulse-chase experiments on GFP-BP80– and GFP-BP80CT–expressing cells. Figure 2A shows that the latter construct chases into the typical vacuolar degradation product at a slower rate than GFP-BP80. The bottom panel represents the percentage of the core fragment in relation to the total signal from the three bands in each lane, calculated from line scanning of phosphor imaging data from two independent pulse-chase experiments. This clearly confirms the difference between the two constructs at each time point of the chase. For instance, after a 16-h chase, the GFP-core fragment accounts for 40% of the total GFP-BP80 protein, and similar progress of processing is only reached after 32 h for the deletion mutant GFP-BP80CT. This suggests that the CT contains positive information for anterograde transport toward a lytic compartment. However, its presence is not absolutely required, confirming the notion that transport to the lytic compartment may be the default for type I membrane spanning proteins (Barrieu and Chrispeels, 1999).

View larger version (29K): [in this window] [in a new window]   Figure 2. The C Terminus of BP80 Promotes Anterograde Transport. (A) Pulse-chase analysis demonstrating that the BP80 CT influences the speed with which GFP-BP80 reaches the lytic vacuole. Protoplasts cotransfected with either GFP-BP80 or GFP-BP80CT plasmids were pulse labeled for 1 h, after which incubations in chase buffer were performed for the periods of time indicated above each lane. Cell extracts were immunoprecipitated with anti-GFP antiserum and used for SDS-PAGE separation. The bottom panel shows the percentage of signal derived from the core fragment (asterisk) in terms of the total signal detected on each chase point for GFP-BP80 (black line) and GFP-BP80CT (gray line). Line scanning of the various chase points from two independent pulse-chase experiments was conducted using AIDA software (Raytest). The signal from the GFP-core band was quantified in arbitrary units and calculated as a percentage of the total surface area from all peaks in each lane. Error bars represent standard error from two independent experiments. Notice that the percentage of GFP-core derived from GFP-BP80 is approximately twofold higher than that for GFP-BP80CT at each time point in the analysis. (B) The absence of the CT compromises acquisition of endoH-resistant glycans. Protoplasts mock transfected (Con) or transfected with 30 µg of either GFP-BP80 or GFP-BP80CT plasmids were incubated for 24 h for the proteins to reach a steady state distribution. Cell extracts were incubated for 1 h in the presence (+) or absence (–) of endoH and verified by protein gel blot analysis. The top two bands represent glycosylated and deglycosylated full-length precursors, and the bottom band represents GFP-core (asterisk). All other annotations are as in Figure 1. Notice a clear shift from the glycosylated to the deglycosylated band with endoH treatment of GFP-BP80CT, indicating partial ER retention of the deletion construct.

To test endoH resistance, protoplasts were transfected with equal amounts of plasmid encoding either GFP-BP80 or GFP-BP80CT. Following transformation, each protoplast suspension was incubated for 24 h, after which protein extracts were prepared and split into two equal aliquots. These were subjected to either mock (–) or endoH (+) digestion. Figure 2B shows that most of the GFP-BP80 molecules are resistant to endoH digestion, indicating that this protein is efficiently exported to the Golgi. In the case of GFP-BP80CT, a larger proportion of the molecules are sensitive to the enzyme, as seen by a clearly reduced signal from the upper band and a corresponding increase in intensity for the deglycosylated polypeptide (Figure 2B, last lane). This suggests that the slower anterograde transport of GFP-BP80CT is, at least to some extent, caused by a less efficient ER-to-Golgi transport. The BP80 CT could thus contain sorting information for ER export, but further experiments are required to identify the putative sorting signal.

The CT of BP80 Directs Increased Anterograde Transport to a Lytic Compartment
We have previously shown that a fusion protein in which the TMD and CT of calnexin were fused to GFP (GFP-CNX) did not cause any disturbance of the BP80-mediated vacuolar transport route (daSilva et al., 2005). Consistently with this observation, GFP-CNX did not produce the vacuolar GFP-core, and its subcellular distribution was unaffected by the drug wortmannin, which is in sharp contrast with GFP-BP80 (daSilva et al., 2005). To investigate how membrane anchoring contributes to the recruitment of sorting machinery that interacts with the CT of BP80, we modified GFP-CNX through replacing the CT of CNX by the corresponding region of BP80, yielding the hybrid molecule GFP-CNX/BP80 (Figure 3A ). If membrane anchoring of the BP80 CT alone is sufficient to facilitate interaction with sorting machinery, GFP-CNX/BP80 should exhibit strong competition activity in our cargo secretion assay.

View larger version (20K): [in this window] [in a new window]   Figure 3. The TMD of BP80 Is Required for Efficient Competition with the Endogenous Receptor. (A) Schematic representation to compare the BP80 and CNX chimeras with their derivative hybrid molecule. The CT of CNX in the noncompetitor GFP-CNX (daSilva et al., 2005) was replaced by the CT of BP80, yielding the hybrid GFP-CNX/BP80. A further control was created from GFP-CNX by deleting the CT (GFP-CNXCT). (B) Competition assay comparing the effects of hybrid GFP-CNX/BP80 and GFP-CNX on the secretion index of amy-spo. Protoplasts were transfected with a constant amount of amy-spo plasmid alone (con) or together with dilution series of either GFP-CNX/BP80 or GFP-CNX plasmids (concentration is shown below each lane). The amy-spo activity was obtained for both medium and cell fractions, and the secretion index was calculated. Standard errors are from four independent protoplast transfections. Notice that the induced secretion index is much lower compared with the GFP-BP80–mediated effect in Figure 1C and that the secretion index is shown at a more sensitive scale. (C) Glycan processing analysis to compare progress through the secretory pathway of the BP80 chimeras. Cell extracts of protoplasts expressing either GFP-BP80CT, GFP-CNX/BP80, or GFP-CNX for 24 h were prepared. Two equal aliquots from each cell extract were subjected to either mock (–) or endoH (+) treatment for 1 h. The resulting samples were separated by SDS-PAGE, followed by immunodetection analysis with anti-GFP antibodies. Mock-transfected protoplasts were used as control for the protein gel blot analysis (con). Further annotations are as in Figures 1 and 2. Notice the lower degree of endoH resistance and increased amount of GFP-core (asterisk) in the GFP-CNX/BP80 lane compared with the GFP-BP80CT lane. (D) Evidence for passage through the Golgi apparatus. Transfected protoplasts producing either GFP-BP80CT or GFP-CNX/BP80 were split in two equal portions and incubated with (+) or without (–) 10 µM BFA. Mock-transfected protoplasts were used as controls for the protein gel blot analysis (con). Notice that GFP-core (asterisk) is only visible in the absence of the drug. (E) Control experiment to show that the presence of BP80 CT, rather than the absence of the CNX CT, causes increased anterograde transport of the hybrid GFP-CNX/BP80. Cells were transfected with constructs encoding either GFP-CNXCT, GFP-CNX, or GFP-CNX/BP80. Annotations are as above. Notice that deletion of the tail from CNX does not yield GFP-core (asterisk) formation.

To test whether the BP80 CT contains ER export information, we compared the endoH sensitivity of GFP-BP80CT with the hybrid GFP-CNX/BP80 and GFP-CNX. Figure 3C shows that the truncated molecule GFP-BP80CT displayed the highest level of endoH resistance. GFP-CNX/BP80 showed only a very small proportion of the endoH resistant form, and GFP-CNX was completely endoH sensitive.

One possible explanation is that the hybrid is exported from the ER faster than GFP-CNX but slower than GFP-BP80CT. Interestingly, increased endoH resistance was not directly proportional to the degree of proteolytic trimming to the core fragment. Despite a higher degree of endoH resistance, GFP-BP80CT showed lower levels of the core fragment. This apparent paradox could be explained if the hybrid molecule reaches the Golgi slower but proceeds to the PVC faster compared with the deletion mutant. This would result in low steady state levels of the hybrid in the Golgi apparatus.

To test that GFP-core formation is due to traffic through the Golgi apparatus in both cases, we tested the effect of the drug brefeldin A (BFA). Figure 3D shows that proteolytic processing to the core fragment is sensitive to the drug and thus rules out that the hybrid bypasses the Golgi apparatus.

One further control experiment was performed to test if the difference between GFP-CNX and GFP-CNX/BP80 was due to the presence of the BP80 CT or due to the absence of the CNX CT itself. This was important because the latter is known to contain a di-Arg motif suggested to mediate ER retention (Barrieu and Chrispeels, 1999). For this purpose, a GFP-CNXCT construct was generated (Figure 3A) to allow a direct comparison with GFP-CNX. Figure 3E shows that deletion of the CT alone does not confer GFP-core formation as seen for the hybrid.

Together, the results strongly suggest that the CNX TMD appears to prevent efficient ER export. Moreover, they indicate that efficient anterograde BP80 sorting is likely to be dependent on both the TMD and the CT of the vacuolar sorting receptor.

Both the TMD and the CT of BP80 Contribute to Efficient Targeting to the PVC
The combination of biochemical assays described above provides strong evidence, but further analysis of individual transport steps from the ER via the Golgi apparatus to the PVC could be complemented by in vivo imaging. To directly observe the intracellular localization of the GFP fusions, they were coexpressed with established YFP protein markers followed by confocal laser scanning microscopy (CLSM). The sialyl-transferase fusion ST-YFP (Brandizzi et al., 2002a) and a PEP12/SYP21 fusion, YFP-PEP12 (Ueda et al., 2004; Uemura et al., 2004), were used to label the Golgi apparatus and the PVC, respectively.

We have previously shown that GFP-BP80 labels punctate structures of variable sizes, which rarely colocalize with the Golgi marker (Figure 6 in daSilva et al., 2005). Here, predominant PVC localization of GFP-BP80 was confirmed by the high degree of colabeling with YFP-Pep12 (Figure 4A ). In contrast with GFP-BP80, GFP-BP80CT is partially retained in the ER and causes increased formation of membrane sheets (Figure 4B). This explains the increased endoH sensitivity. However, the deletion mutant does colocalize significantly with the Golgi marker ST-YFP, although extra-Golgi punctate structures can be detected (Figure 4B, open arrowheads). Consistently, GFP-BP80CT also exhibits a low degree of colocalization with the PVC marker (Figure 4C). The partial accumulation in the Golgi apparatus corresponds with the finding that the deletion mutant still acquires some endoH resistance (Figure 3C). However, in comparison with GFP-BP80, the deletion mutant is significantly retained in the ER, which suggests a role of the CT in ER export.

View larger version (32K): [in this window] [in a new window]   Figure 6. The YMPL Motif Is Crucial for the GFP-BP80 Effect. (A) The concentration-dependent influence of GFP-BP80-Y600A and GFP-BP80-Y612A coexpression on the secretion index of amy-spo. Tobacco mesophyll protoplasts were transfected with a constant amount of amy-spo plasmid alone or together with a dilution series of either GFP-BP80-Y600A or GFP-BP80-Y612A encoding plasmids. Concentrations of plasmid encoding the effector proteins are given below each lane. Standard errors are from five independent protoplast transfections. Note that competition is much reduced for GFP-BP80-Y612A, while the control mutation GFP-BP80-Y600A behaves similarly to wild-type GFP-BP80 (cf. with Figure 1A). (B) Similar to (A) but using maximal concentrations (90 µg) of plasmid encoding either GFP-BP80-Y612A, GFP-BP80CT, or GFP-CNX to allow direct comparison. The amy-spo secretion index is shown with a 10-fold more sensitive scale compared with (A) to appreciate residual competition activity of the Y612A mutant. Standard errors are from five independent protoplast transfections. (C) Total cell extracts from protoplasts expressing either GFP-BP80, GFP-BP80-Y612A, GFP-BP80CT, or GFP-CNX for 24 h were prepared and analyzed via protein gel blots to investigate the comparative ratio of GFP-core fragment for each molecule. Note that GFP-BP80-Y612A gives rise to an intermediate level of GFP-core fragment (asterisk) compared with GFP-BP80 and GFP-BP80CT.

View larger version (58K): [in this window] [in a new window]   Figure 4. Subcellular Localization of the Receptor Chimeras via CLSM. Tobacco leaf protoplasts were cotransfected with 10 µg of plasmids encoding either of the GFP or YFP fusion proteins used in the different experiments below. Cells were incubated for 24 h before analysis. Shown are cross sections through the cell cortex to allow visualization of the cytosol with the intermediate organelles of the secretory pathway. The GFP and YFP fusions are pseudocolored in green and red, respectively. Colocalization is pseudocolored in yellow. Images in bright field (differential interference contrast [DIC]) are shown on the right side. Bars = 10 µm. (A) Protoplast cotransfected with GFP-BP80 and the PVC marker YFP-PEP12–encoding plasmids. Notice that GFP-BP80 almost totally colocalizes with YFP-PEP12, illustrating its predominant localization at the PVC. (B) Protoplasts were transfected with GFP-BP80CT plasmid together with the Golgi marker ST-YFP–encoding plasmid. GFP-BP80CT is noticeably retained in the ER and also accumulates in the Golgi apparatus. The deletion mutant also reaches punctate extra Golgi structures (open arrowheads). (C) Protoplasts were transfected with GFP-BP80CT plasmid together with YFP-PEP12–encoding plasmid. GFP-BP80CT is noticeably retained in the ER and reaches the PVC inefficiently, illustrated by its partial colocalization with YFP-PEP12 (closed arrowhead). (D) and (E) Protoplasts coexpressing the hybrid molecule GFP-CNX/BP80 together with either ST-YFP (D) or YFP-PEP12 (E). Notice that GFP-CNX/BP80 is partially retained in the ER and clearly shows a higher degree of colocalization with the PVC marker than with the Golgi marker. Notice that most punctate signals from GFP-CNX/BP80 are separate from the ST-YFP structures, whereas only few punctuate GFP-CNX/BP80 structures are separate from the PVC marker YFP-pep12 ([E], open arrowheads). (F) and (G) Similar to previous panels, but protoplasts are coexpressing either GFP-CNX (F) or GFP-CNXCT (G) with the Golgi marker ST-YFP. No punctuate structures were observed, and the two GFP constructs are found in ER tubules and sheets.

The frequency of detectable ER labeling versus punctate labeling is shown in Supplemental Table 1 online for the various constructs used in Figure 4. The results strongly suggest that the CNX TMD appears to prevent efficient ER export, while the presence of the BP80 CT stimulates export as well as arrival in the PVC. Statistical analysis of the percentage of colocalization showed that extra ER punctate structures labeled with GFP-CNX/BP80 were mostly colocalized with the PVC marker, similar to GFP-BP80 (see Supplemental Figure 1 online). Although it was hard to detect the hybrid GFP-CNX/BP80 in the Golgi apparatus, Golgi-dependent traffic was shown using the drug BFA (Figure 3D). Together, the results suggest that the BP80 CT confers the ability to compete with endogenous BP80 (Figure 3B) and progress to a lytic compartment yielding GFP-core (Figure 3C) and extra Golgi punctuate structures that represent PVCs (Figures 4D and 4E).

Multiple Sequence Motifs of the BP80 CT Contribute to Receptor Sorting
The cytosolic domain of Arabidopsis BP80a used in this study contains 37 amino acid residues, which is small compared with the long tails of the mannose-6-phosphate receptors in mammals (Ghosh et al., 2003) or VPS10p in yeast (Marcusson et al., 1994). This relatively small size and the high degree of sequence similarity among BP80 isoforms makes it an attractive system to investigate the role of putative cytosolic motifs on its sorting. The BP80 C terminus may contain several sorting signals for the various transport steps, including ER-to-Golgi transport, Golgi-to-PVC transport, and recycling from the PVC. It was expected that interference with any of these transport steps would diminish the ability to compete with endogenous BP80 for sorting machinery. This would manifest itself in a reduced induction of amy-spo secretion, thus forming a simple and reproducible test system to score mutants.

We thus mutagenized all the residues within the tail that are conserved between the known BP80-like molecules in plants (Figure 5A ). Recombinant plasmids were first tested for their ability to produce GFP-BP80 fusion proteins and to monitor processing to the GFP-core fragment (Figure 5B). In a second step, the mutants were screened by scoring the ability to induce amy-spo secretion (Figure 5C), in direct comparison with the highly competitive GFP-BP80 fusion or the noncompetitive deletion mutant (GFP-BP80CT). Figure 5B shows several examples of mutants in which the processing to the characteristic vacuolar GFP-core fragment was either reduced (M601A, D602A, and Y612A) or increased (I608A and L615A). Reduced core formation could be due to impaired anterograde transport or increased recycling. Increased core formation would suggest the opposite: accelerated anterograde transport or defective recycling from the PVC.

View larger version (47K): [in this window] [in a new window]   Figure 5. Identification of Amino Acid Residues on the BP80 CT That Interfere with the Ability to Compete with the Endogenous Receptor. (A) Sequence alignment of the predicted CT on various BP80-like molecules identified in Arabidopsis. Conserved amino acid residues in the tail, highlighted in gray, were individually replaced by an Ala residue via site-directed mutagenesis, as indicated below. (B) Protoplasts were transfected with a constant concentration of amy-spo plasmid alone (con) or together with 30 µg of plasmid encoding either GFP-BP80, the deletion mutant GFP-BP80CT, lacking the entire CT, or one of the various GFP-BP80 mutants (indicated by the usual mention of the position in the protein sequence; BP80 isoform a). Protoplast suspensions were incubated for 24 h, after which cell and medium were harvested. A portion of the protoplasts was extracted for total cellular contents and analyzed by protein gel blotting to compare the ratio between full-length precursors and the lower molecular weight GFP-core fragment (asterisk). After scanning several protein gel blots, reproducible decreases in core/precursor ratio (closed arrows) or increases in the core/precursor ratio (open arrows) were scored. The wild-type control GFP-BP80 exhibited an average of 20% ± 6% GFP-core of the sum of all signals in the lane and was used as a reference. Significantly increased processing was observed for I608A (35% ± 8% core) and L615A (37% ± 7% core), while decreased processing was seen for M601A (1% ± 0.6% core) and D602A (6% ± 3% core). (C) Screening of mutants via the receptor competition assay. Protoplast suspensions in (B) were harvested to obtain cells and medium separately. The amy-spo activity was measured for both medium and cell fractions, and the secretion index was calculated. Standard errors are from three independent protoplast transfections. Secretion index values that were significantly below levels obtained with the positive control GFP-BP80 are indicated in dark gray.

The YXX Motif of the BP80 Tail Is Required for in Vivo Competition with Sorting Machinery
Particularly noteworthy were the point mutations Y612A and L615A that displayed strongly reduced activity in the receptor competition assay (Figure 5C). Interestingly, the two mutations did not show the same effect on the formation of the soluble GFP-core. While the Tyr-to-Ala substitution reduced core formation and suggests reduced leakage to the vacuoles, the Leu-to-Ala mutation increased the leakage to the vacuole, as indicated by an increase in the core signal (Figure 5B).

The Tyr and Leu residues mentioned above are thought to be part of the typical YXX signature of a sorting motif known to interact with the µ-subunit of AP complexes in mammalian cells (Ohno et al., 1998; Owen and Evans, 1998). The data in our plant system strongly suggest that this motif has a biological function in vivo. However, the results also show that the residues at the beginning and the end of the YXX motif may not have the same functions, despite being part of the same perceived consensus sequence. This finding may reveal an unknown complexity of the YXX motif, but for the remainder of this work, we have concentrated on the effects of the Tyr-to-Ala substitution on this motif (mutant Y612A).

Mutation of another Tyr (Y600A), in a region closer to the TMD, had no measurable effect on receptor competition (Figure 5C), illustrating the specificity of the assay. These two Tyr mutants were directly compared in a dose–response experiment. Figure 6A shows that the Y600A mutant essentially behaved like nonmutated GFP-BP80 (cf. with Figure 1), whereas the second mutant (Y612A) only showed weak competition, as indicated by a much reduced induction of amy-spo secretion.

To test the weak competition activity of the Y612A mutant further, we compared it with the complete deletion of the CT (GFP-BP80CT) or the CNX fusion (GFP-CNX) in a dose–response experiment. This analysis confirmed that the Y612A mutant still exhibited a weak but reproducible competition (Figure 6B, note the different scale compared with 6A). Similar levels of the three fusion proteins were produced for the plasmid concentration used (Figure 6C, last three lanes). Interestingly, the levels of GFP-core fragment yielded by the mutant (GFP-BP80-Y612A) are intermediate between the levels resulting from GFP-BP80 and GFP-BP80CT. The results indicate that this Tyr-based motif plays a crucial role in the targeting of BP80 in vivo.

GFP-BP80-Y612A Is Partially Missorted to the Plasma Membrane
To confirm our results with transient expression, we compared the effects of GFP-BP80 with that of the mutant (GFP-BP80-Y612A) in stable transformed tobacco plants. It was noted that transformation with GFP-BP80 yielded many kanamycin-resistant lines that failed to express any detectable amount of recombinant protein. Those lines that expressed the fusion protein did so at low levels and contained mainly the GFP-core fragment resulting from the leakage to the vacuoles. By contrast, transformation with GFP-BP80-Y612A was more frequent and yielded higher expression levels. Figure 7A shows a comparison of selected transgenic lines expressing either GFP-BP80 or the Y612A derivative at similar levels. It is clearly noticeable that the mutant shows a much increased level of the full-length precursor compared with the GFP-core fragment. This confirms the transient expression data on GFP-BP80-Y612A (Figure 6C), suggesting that anterograde transport followed by proteolytic cleavage is reduced but not abolished.

View larger version (54K): [in this window] [in a new window]   Figure 7. Effects of GFP-BP80 and GFP-BP80Y612A Production in Stably Transformed Tobacco Plants. (A) Screening of tobacco plants transformed with either GFP-BP80 or GFP-BP80Y612A construct, resulting in overexpression of the GFP fusion proteins. Extracts were prepared from leaves of untransformed plants (UT) as controls and from independent transgenic plants overexpressing either GFP-BP80 or GFP-BP80-Y612A. Bands of the full-length GFP fusion proteins (open arrowheads) and the corresponding GFP-core degradation product (asterisks) are indicated. Note the increase in the full-length molecule at the expense of the GFP-core in the case of plants expressing the mutant molecule. (B) Transient expression to test the capacity of vacuolar sorting in plants stably producing GFP-BP80 or GFP-BP80Y612A. Leaf protoplast from untransformed (UT) and stably transformed lines obtained from each of the two constructs and expressing similar levels of recombinant proteins were transformed with a constant concentration of plasmids encoding either the secretory marker -amylase (amy) or the BP80-ligand amy-spo. The secretion index is given in different scales for amy (left y axis) and amy-spo (right y axis) to simplify comparison. Standard errors are from three independent protoplast transfections. Note that secretion of the vacuolar cargo amy-spo (gray bars) is induced in protoplasts from GFP-BP80 plants, while amy (white bars) is secreted at comparable levels by the three protoplast populations. (C) Still images from Supplemental Movie 1 online, showing the first (0.000 s), middle (96.483 s), and last (194.936 s) frame. Shown is a guard cell and parts of adjacent epidermis cells analyzed by CLSM in single channel, allowing rapid frame capture. Notice the variable shapes and mobility of the highlighted structures, ranging from punctate to tubular.

The biochemical phenotype of the transgenic plants confirms the competition data with transient expression, showing that GFP-BP80 has an inhibitory effect on BP80-mediated vacuolar sorting and that such a property is greatly compromised by the disruption of the YMPL motif. It should be noted that the observed induction of secretion in the stable transformed lines is 10-fold lower compared with the strongest induction seen with transient cotransfection (i.e., Figure 1C, Y612A). Most likely, high GFP-BP80–expressing transgenic lines exhibiting more pronounced missorting of BP80-ligands were not viable and failed to regenerate. This is not an issue with the transient expression assay, allowing a more sensitive and also more rapid analysis (i.e., Figure 6A).

In leaf epidermis cells, GFP-BP80 labeled highly mobile and morphologically diverse structures. These are punctate, tubular, and sometimes tubular-branched and generally are smaller and faster moving than the typical Golgi bodies (Figure 7C; see Supplemental Movie 1 online). This corresponds well with the size and variable morphology reported for multivesicular bodies of tobacco bright yellow 2 (BY2) cells (Tse et al., 2004). Since stable transformed GFP-BP80–producing plants only exhibited low expression levels, this material may be ideal for further studies on PVC movement.

Next, we compared the intracellular localization of GFP-BP80-Y612A by fluorescence microscopy in transiently expressing protoplasts or in the leaf epidermis of stable transformed lines. GFP-BP80-Y612A was partially retained in the Golgi apparatus, as shown by partial colocalization with the Golgi marker ST-YFP (Figure 8A ). However, the mutant also showed partial colocalization with the coexpressed PVC marker YFP-pep12 (Figure 8B). These results suggest that anterograde transport to the PVC is compromised but not abolished in the Y612A mutant.

View larger version (85K): [in this window] [in a new window]   Figure 8. GFP-BP80Y612A Is Partially Missorted to the Plasma Membrane. (A) and (B) CLSM showing that mutation of the YMPL motif causes changes in subcellular distribution of GFP-BP80. Similar conditions and annotations as in Figure 4, but protoplasts were transfected with plasmid encoding GFP-BP80-Y612A, together with either the Golgi marker ST-YFP (A) or the PVC marker YFP-PEP12 (B). (C) Protoplast cotransfected with GFP-BP80-Y612A and YFP-BP80 plasmids. Shown is a cross section through the center of the protoplast to appreciate labeling of the plasma membrane. A colocalizing signal is indicated by a closed arrowhead, and nonoverlapping signals are indicated by open arrowheads. Notice that only GFP-BP80-Y612A highlights an uninterrupted line surrounding the entire protoplast border and also highlights punctuate structures that are not seen with YFP-BP80. (D) to (F) Leaf epidermal cells of transgenic lines expressing GFP-BP80 (D) and GFP-BP80-Y612A (E). Notice that in addition to punctate structures, epidermal cells of GFP-BP80-Y612A plants show a diffuse staining of the apoplast. This is also seen in guard cells from plants expressing GFP-BP80-Y612A (F). (G) Transgenic BY2 line stably transformed with the GFP-BP80-Y612A construct. Notice again the apoplastic staining. (H) and (I) Cross section through the center of several leaf protoplasts prepared from stable transformed lines expressing GFP-BP80-Y612A. Prior to imaging, protoplasts were incubated for 48 h in TEX medium to allow accumulation of full-length GFP-BP80-Y612A at the plasma membrane. The suspension was then split in two portions and incubated for a further 2 h either with (I) or without (H) wortmannin (10 µM). Notice the drug-induced redistribution of plasma membrane fluorescence to diffuse staining of the vacuolar lumen.

It was also observed that GFP-BP80-Y612A accumulated in additional punctate structures compared with YFP-BP80 (Figure 8C, open arrowheads). This suggests that compromised Golgi-to-PVC transport could lead to missorting to the cell surface and accumulation in different compartments, such as the Golgi apparatus, through which the molecule transits.

Statistical analysis of the percentage of colocalization between either GFP-BP80, GFP-BP80CT, GFP-CNX/BP80, or GFP-BP80-Y612A in combination with either the Golgi marker ST-YFP or the PVC marker YFP-pep12 confirmed a role of the Tyr motif in anterograde transport. GFP-BP80CT and GFP-BP80-Y612A both showed much stronger colocalization with the Golgi marker and a concomitant lower colocalization with the PVC marker (see Supplemental Figure 1 online). By contrast, the constructs carrying the wild-type BP80 CT (GFP-BP80 and GFP-CNX/BP80) exhibited mostly PVC localization and little Golgi retention. The two constructs showed very similar distributions (see Supplemental Figure 1 online) with the exception of the earlier documented ER labeling for the hybrid (Figure 4; see Supplemental Table 1 online). These findings are consistent with the model that both deletion of the tail and point mutation of Y612 lead to reduced Golgi export and, thus, increased Golgi retention.

Analysis of the leaf epidermis of stably transformed tobacco lines also revealed a clear difference in the localization of GFP-BP80 and the mutant protein. Figure 8D shows an overview of the fluorescence in the epidermis. In contrast with GFP-BP80, GFP-BP80-Y612A–producing plants showed diffuse apoplastic GFP fluorescence and intracellular punctate structures, confirming a significant amount of missorting to the apoplast, followed by cleavage of GFP (Figures 8E and 8F). Stable transformed tobacco BY2 suspension cultures showed similar apoplastic fluorescence with the mutant (Figure 8G). This was not observed for the GFP fusion carrying the wild-type BP80 tail (data not shown).

To avoid proteolysis of GFP at the plasma membrane, we took advantage of our experience with protoplast suspensions from transgenic plants. Due to the dilution, the medium of washed protoplasts is less lytic compared with the concentrated apoplast fluid of leaf epidermis cells. When protoplasts were prepared from the leaves of transgenic plants, washed repeatedly, and then incubated in fresh medium, plasma membrane fluorescence increased gradually over time and reached maximum levels after 48 h in the case of cells expressing GFP-BP80-Y612A (Figure 8H). Moreover, no soluble cleaved GFP-core fragment was seen in the culture medium (data not shown). This confirms that GFP-BP80-Y612A reaches the plasma membrane as an intact membrane protein but is cleaved in the apoplast of intact tissues. The diluted environment of the protoplast incubation medium does not appear to be sufficiently lytic to mediate cleavage and thus reveals traffic to the plasma membrane.

Wortmannin-Induced Rapid Redistribution of Plasma Membrane–Localized GFP-BP80-Y612A to the Vacuole Suggests an Endocytic Route for the Receptor
Missorting to the plasma membrane could be explained by a variety of scenarios. One could assume that BP80 normally traffics via the plasma membrane to reach the PVC by endocytosis and that this is disrupted by the Y612A mutation. However, it was shown that the YXXL motif of BP80 interacts in vitro with the µ-subunit of mammalian AP1, normally associated with budding of clathrin-coated vesicles from the Golgi apparatus (Sanderfoot et al., 1998), but did not interact with the equivalent subunit of AP2 (associated with clathrin-coated vesicles budding from the plasma membrane). More recently, the YXXL motif was shown to interact with a µ-subunit of Arabidopsis that localized to the Golgi apparatus (Happel et al., 2004). The current state of the field therefore favors a model in which normal BP80 traffic occurs from the Golgi apparatus directly to the PVC and depends on information within the YMPL motif.

Here, we have observed that the Y612A mutation results in mistargeting to the plasma membrane and inefficient progress to the PVC. There are at least three explanations for this observation. One possibility is the existence of a clathrin-independent Golgi-to-PVC route in plants that is slower but can be used as an alternative. This could be the previously proposed default pathway for membrane proteins (Barrieu and Chrispeels, 1999). It is also possible that the clathrin-mediated route may carry bulk flow in addition to selectively recruited cargo. A third explanation could be provided by recent findings with the yeast Saccharomyces cerevisiae. In the absence of the direct clathrin-mediated Golgi-to-PVC route, the yeast vacuolar sorting receptor VPS10 was shown to reach the PVC by a detour via the plasma membrane (Deloche and Schekman, 2002), a route not normally followed by this receptor.

To test these possibilities experimentally, we took advantage of our previous observation that the drug wortmannin is a strong inhibitor of the route taken by BP80 to recycle from the PVC to the Golgi apparatus (daSilva et al., 2005). If the first two models are correct, arrival of GFP-BP80-Y612A at the plasma membrane would result from saturation of the nonselective Golgi-to-PVC route. In that case, plasma membrane fluorescence would not be influenced by wortmannin because wortmannin does not affect the constitutive ER-to-Golgi plasma membrane route (Pimpl et al., 2003). If the third model is correct, endocytosed GFP-BP80-Y612A would not recycle from the PVC to the Golgi apparatus in the presence of the drug. This would lead to accumulation of the fusion protein in the vacuole at the expense of the amount in the plasma membrane.

Therefore, leaf protoplasts from the GFP-BP80-Y612A–expressing transgenic plants were prepared and incubated in protoplast culture medium for 48 h to reach easily detectable steady state levels of the fusion protein at the plasma membrane. The sample was then split into two equal aliquots, and one of which was supplemented with wortmannin to a 10 µM final concentration. Both samples were incubated for up to 4 h, and GFP fluorescence was compared through CLSM. Drug treatment caused a rapid reduction in the fluorescence at the plasma membrane, which was accompanied by an increase in the labeling of the central vacuole lumen (Figure 8I). This redistribution is rapid and involves the majority of the plasma membrane fluorescence that required 48 h to build up. It is therefore unlikely that increased vacuolar fluorescence is due to de novo synthesis.

The results suggest that BP80 carries signals for endocytosis that may be used when BP80 reaches the plasma membrane. These must still be present in the Y612A mutant and may be found by analyzing individual point mutations found in our screen (Figure 5). It may be necessary to combine individual mutations to identify the corresponding signal.

The YXX Motif of the BP80 Tail Is Required for in Vivo BP80 Function
To substantiate the data obtained with the GFP fusion proteins, we tested the influence of the Tyr motif on the natural Arabidopsis receptor BP80a. We thus created the Y612A mutant within the context of the entire BP80 coding region and tested the ability to reconstitute vacuolar sorting of the BP80-ligand amy-spo in the presence of GFP-BP80 as competitor. It was shown previously that excess of wild-type receptor can rescue vacuolar sorting under these conditions, providing a positive assay for receptor function (daSilva et al., 2005). Figure 9A confirms that wild-type BP80 coexpression can overcome competition by GFP-BP80 (gray bars). By contrast, BP80-Y612A aggravated induced secretion of amy-spo (Figure 9A, white bars).

View larger version (20K): [in this window] [in a new window]   Figure 9. Tyr-Based Motif of Full-Length BP80 Is Required to Alleviate GFP-BP80 Effect. (A) Transient expression experiment to reconstitute vacuolar sorting with full-length BP80. Protoplasts were electroporated with a constant concentration of amy-spo plasmid, either alone (black bar, first lane), with GFP-BP80 (black bar, second lane), or with GFP-BP80 supplemented with a dilution series of either wild-type BP80 (gray bars) or mutant BP80 (Y612A) (white bars). Concentrations of GFP-BP80 and the two BP80 constructs are indicated below each lane. Standard errors are from four independent protoplast transfections. Notice that wild-type BP80 reconstitutes vacuolar sorting as seen by a reduced secretion index, while the Tyr mutant aggravates the effect of competitor GFP-BP80 and increases amy-spo secretion even further. (B) Overexpression of Arabidopsis full-length BP80 and its Y612A mutant in stable transformed tobacco. Total leaf extracts from one untransformed plant (UT), two BP80 (BP80), and two BP80Y612A mutant (BP80Y612A) overproducing transgenic lines were compared directly (left panel). Proteins were separated by SDS-PAGE followed by gel blotting analysis with polyclonal anti-BP80 serum anti-VSRAt-1 (Tse et al., 2004). The right panel shows gel blotting analysis of total cell extracts from protoplasts prepared from the same plant material shown in the left panel. Notice that the BP80-Y612A overproducers yield several lower molecular weight degradation products, which are not detectable in protoplasts from the same material. (C) Leaf protoplasts prepared from untransformed tobacco (UT), BP80, and BP80-Y612A plants were transfected with a constant amount of plasmid encoding the cargo vacuolar cargo amy-spo on its own (–) or together (+) with a constant concentration of GFP-BP80 plasmid. Standard errors are from three independent protoplast transfections. Notice that protoplasts from BP80-Y612A plants exhibit a comparatively higher secretion index for amy-spo, which is further increased by coexpression of GFP-BP80.

Figure 9B shows leaf extracts from transgenic lines expressing the two recombinant proteins at comparable levels. Interestingly, the extracts from the plants expressing the mutant BP80 exhibited characteristic lower molecular weight degradation products (Figure 9B, left panel, compare BP80 with BP80-Y612A). These were water-soluble and could only be detected in extracts from entire leaf tissues but not isolated protoplasts (Figure 9B, right panel). This suggests that proteolytic cleavage did not occur intracellularly and originated from apoplastic proteases. The results confirm that in the absence of a functional YXX motif, a portion of the mutant receptor reaches the plasma membrane, where a lower pH and secreted proteases could cause cleavage of the receptor domain in plant tissues. This corresponds well to the diffuse apoplast staining observed for GFP-BP80-Y612A.

To test BP80 targeting to the apoplast in a functional way, we analyzed the transport properties of the BP80-ligand amy-spo in protoplasts prepared from either untransformed plants, wild-type BP80, or BP80-Y612A–producing plants. The different protoplast suspensions were transfected with plasmid encoding the cargo amy-spo in the presence or the absence of the competitor GFP-BP80. Figure 9C shows that BP80-overproducing plants exhibited increased resistance to the competitor GFP-BP80, as seen by a reduced induction of amy-spo secretion (cf. lane 2 with 4, white bars). However, mutant BP80 expression leads to a strongly induced amy-spo secretion under control conditions (cf. lane 1 with 5, gray bars) and aggravated the effect of the competitor GFP-BP80, as seen in Figure 9A in the transient assay.

The biochemical phenotype of the transgenic plants indicates that mutant BP80 binds to amy-spo within the secretory pathway and releases a portion of them after reaching the plasma membrane, probably due to the lower pH of the medium. Transport of the control cargo amy was not affected by either of the transgene products in experiments of Figures 9A and 9C (data not shown), confirming that BP80 affects only its ligands and does not change the fate of constitutively secreted proteins.

The confirmation of the findings with the native receptor strongly support all data obtained with the GFP fusions and illustrate that receptor sorting information is mainly restricted to the TMD and the CT of the receptor. Besides the typical YXX signature that is thought to be responsible for clathrin-mediated anterograde Golgi-to-PVC transport (Happel et al., 2004), multiple sequence motifs are likely to contribute to receptor sorting. Among numerous arguments, this was most convincingly supported by the fact that residual competition was observed with either Y612A or L615A mutants that were well above control levels or the effect of the complete deletion of the tail (Figure 5C). Further signals may control steps as diverse as ER export, endocytosis, and receptor recycling from the PVC and leave plenty of work for the future.

	DISCUSSION

TOP ABSTRACT INTRODUCTION RESULTS DISCUSSION METHODS REFERENCES

 
Receptor Competition: A Powerful Tool to Study Putative Sorting Signals in Vivo
In previous work, we established that a truncated BP80 molecule lacking its ligand binding domain could act as a potent inhibitor of vacuolar sorting by interfering with endogenous BP80 sorting (daSilva et al., 2005). The effect was semidominant because it could be fully reconstituted by wild-type BP80 overexpression. The truncated molecule contained the TMD and CT of BP80, but the lumenal domain was replaced by GFP. It was concluded that sorting information for the receptor must be localized within those two domains. Here, we have systematically explored this quantitative competition assay to study the effects of modifications in those two domains.

It could be established that competition is fully prevented when the entire BP80 CT is removed (GFP-BP80CT). Figure 1 shows that in contrast with GFP-BP80, the deletion mutant sustains no measurable competence to increase the secretion of the vacuolar reporter amy-spo. These data are consistent with the notion that the CT carries information necessary for the recruitment and incorporation of receptors into transport carriers. Such a feature is similar for the mammalian mannose-6-phosphate receptors, which have been shown to contain multiple motifs on their cytosolic domain that interact with components of sorting machinery in the cytosol (Honing et al., 1997; Diaz and Pfeffer, 1998; Wan et al., 1998; Puertollano et al., 2001; Ghosh et al., 2003; Ghosh and Kornfeld, 2004).

We could also show that cytosolic display of the CT alone is not sufficient for titration of sorting machinery and does not compromise BP80-mediated vacuolar sorting (Figure 1). Moreover, display on the cytosolic face of a membrane surface alone is no guarantee for efficient competition either. We have created a fusion protein consisting of GFP, followed by the TMD of CNX and the CT of BP80. The resulting hybrid GFP-CNX/BP80 acts only as a weak competitor and induces the secretion of the vacuolar reporter amy-spo 10-fold lower compared with the original GFP-BP80 fusion (Figure 3). The weak effect is due to an efficient ER retention of the hybrid molecule mediated by the CNX TMD.

GFP fusions carrying the CNX TMD with or without its CT are retained in the ER and do not reach a lytic compartment (Figure 3E). However, the hybrid GFP-CNX/BP80 reaches a lytic compartment. This occurs via traffic through the Golgi apparatus because GFP-core formation was shown to be sensitive to BFA (Figure 3D). The portion of the hybrid that escapes the ER proceeds further and competes with endogenous BP80, leading to weak but measurable competition (Figure 3B). Together, the results strongly suggest that display of the BP80 CT at the Golgi apparatus is necessary to compete with endogenous BP80 for sorting machinery. This is consistent with the current concept that recruitment of specific sorting machinery components is dependent on the recognition of surface determinants characteristic for individual membrane subdomains (Munro, 2004).

The results also provide insight into the retention of CNX. While it was shown that the CT of CNX contains a di-Arg motif that contributes to retention (Barrieu and Chrispeels, 1999), our data indicate that the TMD of CNX also plays a role in reducing ER export. The two findings are not mutually exclusive, and it is possible that CNX uses both true retention and signal-mediated retrieval to reach high steady state levels in ER membranes. We cannot completely rule out that malfolding of hybrid constructs contributes to our findings, but we consider it very unlikely because all constructs were fluorescent. In addition, the export of two independent constructs (GFP-BP80CT and GFP-CNXCT) could be stimulated by the addition of the BP80 CT. Likewise, the presence of the CNX TMD consistently reduced the anterograde transport (cf. GFP-BP80 with GFP-CNX/BP80 and GFP-BP80CT with GFP-CNXCT). It is therefore more plausible that the CNX TMD reduces export, while the BP80CT promotes it. The mechanism by which the CNX TMD triggers the ER retention of GFP-CNX/BP80 is not within the scope of this article, but it does not seem to be merely dependent on its length and calls for further investigation.

Although it was proposed that the competition would mainly occur at the limiting recycling step from the PVC (daSilva et al., 2005), the results obtained suggested that interference with any step in the traffic of the competitor between the ER, the Golgi apparatus, the PVC, and its putative recycling route would manifest itself in a reduced ability to compete wit